Kevin M. Johnson, PhD

University of Wisconsin – Madison

kmjohnson3@wisc.edu

Students, scientists, physicists, engineers, and clinicians interested current and prospective methods for the evaluation of acute and chronic intracranial vascular disease.

This talk aims to provide insights into solutions for current challenges and elicit thoughts for potential changes in the neurovascular imaging landscape. At the end of this talk, participants should:

-Be able to identify limitations of the current MR vascular imaging paradigm and how these techniques compare to state-of-the-art techniques across modalities

-Be prepared to evaluate new vascular imaging techniques and results given in papers presented in following sessions.

Excluding limitations due to logistical and economic factors, MRI is a superior method for comprehensive assessment of the brain. This has been largely spurred by the diverse contrast mechanisms available to probe pathology. In this lecture, we explore the developing methods to assess the neurovascular structures at risk of hemorrhage, culpable vascular disease, and post treatment evaluation.The main goals of hemorrhagic vascular imaging are to predict likelihood of rupture (e.g rupture risk), evaluate the culprit pathology, and longitudinally monitor treatment response. Much of the current clinical decision making for vascular lesions is based on “lumenographic” techniques that depict the lumen of vessels and veins. From these images and prior studies relating geometry and filling patterns to prognosis, treatment actions are considered. The parameters required depend on the disease in question. For aneurysms, this includes the size, location, growth, and appearance of blebs; requiring high spatial resolution (much greater than 1mm). For arteriovenous malformations (AVMs), this includes the identification of filling and draining pattern; requiring high temporal resolution ( less than 1s ). In providing these angiographic metrics universally, X-ray digital subtraction angiography (DSA) is the clear gold standard. Modern DSA systems provide high frame rate dynamic images (>10 fps) and 3D or 4D images with exceptionally high spatial resolution (< 0.25mm). This impressive performance is in addition to vessel selective capacities. State-of-the-art, computed tomography (CT) offers similar features to DSA with the additional ability to quantify perfusion and allow for intravenous injections. Current MRI techniques do not provide the same level of lumenographic detail and thus can miss important details required for surgical planning or diagnosis. Common occurrences include overestimation of stenosis percentage, poor visualization of small aneurysms structures, and incomplete visualization of arteriovenous malformation filling and draining. However, the complimentary information provided by anatomical and functional measures and a minimal safety risk make them viable alternative or preferred in non-acute settings.

Inflow based angiography is currently the most commonly used technique for the assessment of intracranial vessels, predominantly time-of-flight (TOF). TOF is a simple T1 weighted sequence optimized to maximize an existing inflow based contrast. With the development of multi-slab excitation[1], magnetization transfer [2, 3], and use of 3T (or higher) scanners TOF has improved dramatically over the years. In many cases, current product TOF can be used to produce exquisite depiction of the intracranial vascular in 4-8 minutes. However, time of flight MRA remains a lengthy scan in the context of a complete neuro imaging protocol. Significant time savings may be afforded by the introduction of advanced reconstruction and acquisition techniques including, compressed sensing [4] [5] and simultaneous multi slab technologies [6]. With these techniques an additional acceleration factor of 2 may be practical for routine imaging. Despite advances, TOF is insensitive to slow flowing blood. To examine this, please define the time-of-arrival (TOA) as the time required for blood to enter the slab and reach a given location. Given the strong background signal from grey and white matter, a TOF sequence is only sensitive to vessels with TOA less than ~500ms. That is if it takes longer than 500ms to travel from outside to slab, the signal from background signal will be higher than that of blood. This poses a significant problem if examining aneurysms, stenosis, and AVMs which often have slower flowing components of interest due to recirculation zones and venous drainage. The use of 7T simultaneously lengthens the blood T1 and boosts SNR such that the contrast to noise between inflowing blood and static tissue is substantially improved for both slow flowing and fast flowing structures [7] [8]. Unfortunately, high field scanners are not widely available and significant technical challenges remain.

At 1.5 and 3T, recent advances have revolved around techniques to provide reduced sensitivity to blood flow and potential to provide dynamic information, often cast as more formal targeting of blood with Arterial Spin Labeling (ASL) techniques. In this imaging paradigm, blood is specifically tagged during a preparation module and subsequently imaged. This decoupling of imaging and preparation allows background free intracranial MRA via subtraction of images collected with and without the preparation module. Thus unlike TOF, ASL imaging techniques are solely limited by noise and it is relatively easy to produce images of vascular structure with TOA less than 3s. With this substantial reduction in flow sensitivity, single slab inflow MRA images are feasible [9, 10] with better depiction of anatomy with complex or tortuous flow. Of perhaps greater interest, by taking multiple images with different tagging durations, time resolved images can be created [10-14]. Unlike contrast enhanced MRA and CTA techniques, these dynamic images can be acquired with better than 100ms resolution and are not subject to bolus dispersion associated with intravenous injection. Furthermore, modifications to the pulse sequence allow vessel selective tagging [11], bringing complete feature set of DSA to MRA. Alternatively, tagging can be performed with velocity selective-ASL [15], which is insensitive to arrival time and can be used to image arterial and venous structures simultaneously. Decoupling the readout, additionally allows the use of ultra-short echo time techniques, which are more resilient to complex flow patterns and artifacts introduced by metallic devices [16, 17].

When Contrast Enhanced MRA (CE-MRA) was introduced; it made incredible inroads to almost every vascular territory is by far the most dominant method for MR angiography outside the head. This is largely due to improved scan efficiency and greatly improved insensitive to errors related to slow filling vessels. The use of CE-MRA in the intracranial vasculature is unfortunately far more challenging and is thus heavily reliant on high resolution images acquired with alternative techniques. This is namely due to the demand for both high temporal and spatial resolution. A temporal resolution of 1s with 0.5mm spatial would be sufficient to detect most abnormal filling patterns and prevent most significant errors from venous and perfusion overlap. However, this leads to an estimated required acceleration >100x. This is far greater than what is achievable with common acceleration techniques such as parallel imaging. Thus for the past decade, CE-MRA techniques have either ignored dynamic information entirely or relied on clever schemes to accelerate image acquisition (i.e. TRICKS [18],CAPR [19], TWIST, etc). Subsequent CE-MRA images often have substantial artifacts, most often at vessel edges, which must be carefully interpreted to prevent misdiagnosis. With recent developments in reconstruction algorithms, an incredible opportunity exists to more explicitly harness assumption regarding CE-MRA. These techniques, which will often be labeled “compressed sensing” and “low rank approximation”, provide opportunities to provide substantially higher accelerations and/or reduced imaging artifacts[20, 21] [22]. All of these techniques exploit known assumption about the underlying structure (i.e. few vessels, similar temporal dynamics). With these techniques, required acceleration factors of 100x may be possible in the near future. It is important to note that while lumenography is the most dominant method of assessing large vessel vascular disease, many prevalent diseases are due to abnormal endothelial response and subsequent remodeling. Thus alternative measures of the vessel wall health may be of much greater prognostic value than lumenographic techniques alone.Previously relegated to high SNR imaging scenarios in the neck, largely the imaging of atherosclerotic plaques; intracranial vessel wall imaging has made strides towards becoming a viable tool for clinical imaging. Most prominently, 3D variable flip angle fast spin echo imaging has become more widely available [23]. Variable flip angle refocusing greatly increases the sensitivity to flow leading to black blood images. Further, recently developed modules can be incorporated into these sequences to improve flow suppression in cases of slow or recirculating flow. These include motion-sensitized driven equilibrium (MSDE) [24] [25] and delay alternating with nutation for tailored excitation DANTE [26-29]. MSDE creates a velocity sensitive saturation by inserting gradients between a 90 and -90 degree excitations. DANTE, used also for myocardial tagging, creates a series of subvoxel tags which saturate blood as it moves through them. Black blood imaging is a particular promising technique to assess atherosclerotic plaques, but also is potential technique for the imaging of aneurysms. Recent studies of aneurysms demonstrate a correlation between rupture and signals seen in black blood MRI [30] [31, 32]. These signals are generally the post-contrast enhancement seen after injection with an exogenous agent. Currently, the literature has used Gd based agents but also those based on superparamagnetic iron oxide agents (SPIO). Literature suggests the SPIO agents are targeted towards inflammation [33] [34] while the Gd agents may indicate thrombus, permeability, or other non-specific uptake. “Flow Imaging”: The vascular endothelium is sensitive to flow conditions, due to the residence time of locally produced factors and local forces on the wall itself. For example, endothelial cells align with the flow direction and abnormal flow leads derangement. This is especially true for aneurysms which have been shown to have flow sensitive growth and rupture risk [35]. With recent advances flow fields can now be directly probed with 4D Flow MRI ( ecg gated, 3D phase contrast)[36]. This provides a new opportunity to more directly measure the local hemodynamic conditions and how they may be altered to either reduce disease. The interpretation of this information is still highly speculative and the formation of AVM and Anuerysms, still poorly understood. “Perfusion –Contrast Kinetics”: Easier to interpret, perfusion imaging allows a more direct visualization of downstream vascular effects and blood-brain barrier status. Perfusion detects are well characterized with dynamic susceptibility contrast (DSC). However, DSC signals are disrupted in cases of a compromised blood brain barrier. Of greater interest in the case of neurovascular hemorrhage, dynamic contrast enhanced (DCE) perfusion offers the ability to separate and quantify the exchange of agents between the extravascular and intravascular compartments. DCE is identical to high frame rate CE-MRA and is similarly benefiting from advanced acceleration techniques [37]. After imaging, the dynamic data is fit to a two-site exchange model. This has the obvious benefit of being able to image blood barrier disruption due to hemorrhage. Additionally, recent evidence suggests DCE is capable of imaging subtle extravasation of contrast in aneurysms [38], which corroborates enhancement seen with post-Gd vessel wall imaging. A growing also of tools are becoming available for multi-contrast vascular assessment. While it may become increasingly more challenging to compete with x-ray based angiographic techniques; MRI provides so much information beyond the vessel lumen. This is especially relevant as “lumenographic” techniques may become insufficient to grade vascular lesions with advances in pharmaceutical treatment. Here MRI holds potential to become a dominant and comprehensive technique, providing proxy measures of vessel wall health and status of the downstream parenchyma.